I'd love to share this with family and friends to replace my usual incoherent shreiking, but I just happen to be familiar with one of the referenced papers and I really don't think a layperson would agree it even involves AI, much less depends on it: the paper arguing for linguistic characteristics in sperm whales, which would make them the third ever species observed to possibly use language. [1]
They use "custom MatLAB scripts" and certainly employ statistics, but, IDK... seems like a stretch. Actually wow they published their code to GitHub (TIL!), and the closest thing to "AI" is `scipy` for the main scripts.[2]
It's a well-made site and the idea is honestly commendable just from the perspective that so, so many well-meaning people are completely wrong about such important empirical facts.[3] But if there's even one paper that made it through Claude's watchful gaze that I couldn't defend myself as AI, I can't use it.
For the author, some unprompted advice:
1. I think this paper made it through because the lab involved has "Artificial Intelligence" in their name, so it appears throughout the press release your pipeline found. For this to work I think you'll need to target """"generative AI"""" in particular (at least some sort of DL, and ideally transformers or diffusers specifically). I also find this annoying and unscientific (poor GOFAI...), but it's how people understand the term colloquially.
2. The link seems to be to some press release, which sorta implies you're crawling random search results rather than academic sources in particular. Might help!
3. I'd strongly suggest you drop the quasipolemics when you discuss this, especially any mention of datacenters whatsoever. For this to work it's gotta stand on its own, IMHO.
4. Probably should try not to comment on your HN post -- they downrank posts that get comments :))))
I'm hearing a lot about Data Centers Bad and how AI is Taking Jobs™, but I was curious — are we actually getting anything _real_ from AI yet? So I built whatthefuckhasaidone.com. Check it out.
I'd love to share this with family and friends to replace my usual incoherent shreiking, but I just happen to be familiar with one of the referenced papers and I really don't think a layperson would agree it even involves AI, much less depends on it: the paper arguing for linguistic characteristics in sperm whales, which would make them the third ever species observed to possibly use language. [1]
They use "custom MatLAB scripts" and certainly employ statistics, but, IDK... seems like a stretch. Actually wow they published their code to GitHub (TIL!), and the closest thing to "AI" is `scipy` for the main scripts.[2]
It's a well-made site and the idea is honestly commendable just from the perspective that so, so many well-meaning people are completely wrong about such important empirical facts.[3] But if there's even one paper that made it through Claude's watchful gaze that I couldn't defend myself as AI, I can't use it.
For the author, some unprompted advice:
1. I think this paper made it through because the lab involved has "Artificial Intelligence" in their name, so it appears throughout the press release your pipeline found. For this to work I think you'll need to target """"generative AI"""" in particular (at least some sort of DL, and ideally transformers or diffusers specifically). I also find this annoying and unscientific (poor GOFAI...), but it's how people understand the term colloquially.
2. The link seems to be to some press release, which sorta implies you're crawling random search results rather than academic sources in particular. Might help!
3. I'd strongly suggest you drop the quasipolemics when you discuss this, especially any mention of datacenters whatsoever. For this to work it's gotta stand on its own, IMHO.
4. Probably should try not to comment on your HN post -- they downrank posts that get comments :))))
---
[1]: https://www.nature.com/articles/s41467-024-47221-8 Note that language != communication.
[2]: https://github.com/pratyushasharma/sw-combinatoriality/blob/...
I'm hearing a lot about Data Centers Bad and how AI is Taking Jobs™, but I was curious — are we actually getting anything _real_ from AI yet? So I built whatthefuckhasaidone.com. Check it out.